CP620, Shock Compression of Condensed Matter - 2001
edited by M. D. Furnish, N. N. Thadhani, and Y. Horie
© 2002 American Institute of Physics 0-7354-0068-7/02/$ 19.00
DIGITAL SPECKLE X-RAY FLASH PHOTOGRAPHY
S.G. Grantham, W.G. Proud
Cavendish Laboratory, Madingley Road, Cambridge, CB3 OHE. UK
Abstract. The new technique of digital speckle X-ray flash photography (DSXFP), which has been
successfully applied to polyester and cement specimens1, is being further developed and used to study
materials in ballistic situations in a way not previously possible. The technique involves seeding the
specimen with a lead layer and then taking flash X-ray images before and during an impact event.
Digital cross-correlation can then be used to make measurements of the internal displacements occurring
throughout the specimen. Using a stereoscopic geometry the out of plane displacements can also be
determined and a full 3-dimensional displacement map constructed. In this paper these two powerful
and complementary techniques of flash X-rays and DSXFP are used to study the ballistic response of a
borosilicate sample to produce information that other techniques are unable to provide.
INTRODUCTION
per experiment. This has been shown to be the case
in previous work2, however, and hence this
technique can be applied with reasonable
confidence. This still does not allow displacement
measurements within the sample to be made. A
dramatic improvement on flash X-ray photography
is digital speckle X-ray flash photography
(DSXFP). This is a relatively new technique which
allows quantitative measurements of the
displacement on a plane within the sample during a
ballistic impact to be made. The technique requires
a random sprinkling of an X-ray opaque material on
a flat plane within the sample. A "before" image is
taken, and then, another flash X-ray is taken at a
given point during the impact event. The resulting
random speckle patterns can then be correlated to
find the maximum of correlation and hence where
the random pattern has moved to3. By repeating
this correlation using small subimages from the
reference and deformed images, a map of
displacement vectors can be calculated.
The
resolution of these vectors, and hence the resolution
of the displacement measurements, is defined by the
size of these subimages and the stepsize taken
between subimages.
In the following research some traditional flash
X-ray images, of borosilicate glass being impacted,
The study of how glass fractures during a
ballistic impact is of great interest given the
conditions in which glass is often used. The way a
glass windscreen or bullet proof glass reacts to a
high velocity impact are examples of this. In this
paper the processes by which a sample of glass
fractures and fails under such a high velocity impact
is studied.
Techniques that can be used for investigating
the fragmentation of glass in high-speed ballistic
events are limited.
High-speed photography
becomes ineffective when trying to look at the
behaviour behind the damage front, as the glass
comminutes and shatters leading to optical opacity.
Thus, high-speed photography can give the velocity
of the damage front but little else. The use of
gauges is also not a viable option since this involves
altering the structure of the sample quite
dramatically by cutting it and inserting the gauge.
If flash X-ray photography is used this can provide
information on the position of the projectile without
the opacity behind the damage front being a
problem. This technique relies on the sample
fracturing in the same manner in different
experiments, as only one flash X-ray can be taken
803
have been taken which provide information on the
material's failure. A specially manufactured sample
is then used for DSXFP, and a comparison between
these different results made.
configuration used
illustrated in Fig. 1.
is
Fig. 2. through to Fig. 5. show the results from
normal flash X-ray photographs of the mild steel
rod impacting the block of borosilicate glass. It can
be seen in Fig. 1. that the steel rod has been
flattened on impact and has caused cracks to
propagate away from the point of impact and into
the glass. A crack also appears to be forming
directly ahead of the projectile on the rear surface of
the block. Fig. 2. shows the sample beginning to
bulge outwards at the rear surface. The first signs
of debris being ejected from this region can also be
seen. The notch on this projectile is a marker which
was placed 15 mm from the tip of the projectile to
allow the degree of deformation in the rod to be
gauged. At 105 jis following impact (Fig. 4), the
rear surface of the glass is clearly starting to shatter
and break into pieces. However the two sides of the
borosilicate block still appear to be relatively intact.
Whilst they are moving outwards and away from
the projectile, the sides are not suffering from the
same degree of fracturing as the main central
section of the sample. The final X-ray image (Fig.
5), taken at 184 jis post impact, clearly depicts very
little coherent structure left in the sample, whilst the
side sections are also beginning to fail
catastrophically. From these results, the expected
behaviour of the glass in this velocity range has
been observed. When performing the experiment
on the tungsten seeded glass sample, a quantitative
comparison of the structural response can be made.
The X-ray image taken during the impact is
reproduced in Fig. 6. The image is a computer scan
of a contact print taken from the X-ray film
negative. A contact print has been used, rather than
just a scan of the negative, so that more contrast can
be brought up on the image, which allows a better
correlation to be achieved. Two bolts can be seen
behind the sample, these were used to hold the
sample in place and the area behind these bolts at
the top of the picture is an area of fiducial markers.
These were attached to the platform that the
specimen was mounted on so that rigid body
motions can be eliminated. The displacement
Blocks of borosilicate with a 30 x 15 mm cross
sectional area and a length of 60 mm were used for
the standard flash X-ray photographs.
The
projectiles used were mild steel rods 9.15 ± 0.05
mm in diameter and 80 ± 0.5 mm in length with a
rounded tip and were fired at a velocity of 190 ± 3
m s"1. The X-ray images were taken at delays of
26^is, 79|is, 105(is and 184jis using a 150keV X-ray
head with a 30 ns exposure. The X-ray film was
sandwiched in a cassette between two image
intensifier plates to increase the exposure and with a
2 mm lead sheet behind to prevent X-rays escaping
the apparatus. The sample used for the X-ray
speckle study was made by spinning a tube of glass
out into two flat discs, which were then flattened.
These two discs were placed in a carbon holder
with a layer of tungsten filings (50 to 250 (im in
size) sprinkled between them and heated to 950°C
for 1 hour. The sample was then cut to 42 x 42
mm2 and 10 mm thick. The tungsten layer was at a
height of 5 mm through the centre of the sample.
Tungsten was used for this particular experiment
instead of lead because it has a thermal expansion
coefficient that is better matched to glass. This
prevents any cracking of the sample during cooling.
C-RAY!
Tungsten Layer
/
experiments
RESULTS
EXPERIMENTAL
Velocity
for these
BorosiiicE
Sample
FIGURE 1. Experimental setup for DSXFP
The projectile had a velocity of 197.6 ± 3.0 m s"1
and a delay of 40 jis was used. The experimental
804
FIGURE 5. Rod impacting glass at 190 m s"1, 184 {is delay.
1
FIGURE 2. Rod impacting glass at 190 m s" , 26 jis delay.
vectors, which are overlayed, have been scaled up
by a factor of 3 and one pixel represents 0.063 mm
in the frame of the sample. Cracks in the glass are
beginning to propagate away from the tip
(represented by white dashed lines in Fig. 6.) of the
projectile and, again the projectile has been
flattened on impact. The displacement vectors,
produced by the correlation process, appear to agree
with the behaviour previously observed in the
normal X-ray impacts. The central section appears
to be moving backwards away from the projectile in
a large section, and the regions to either side of the
projectile appear to be moving outwards. These are
moving to a lesser degree than the main central
section however. The major fragments of the glass
were retrieved after the experiment and it was found
that out of 154 fragments, only 10 appeared to have
tungsten filings on any of the outer surfaces. From
this it would seem reasonable to infer that the glass
has not preferentially fractured along the seeded
layer.
FIGURE 3. Rod impacting glass at 190 m s"1, 79 jis delay.
CONCLUSION
We have shown that the displacement field, on a flat
plane, within a glass sample can be measured
during an impact event. We have also carried out
standard impact tests on a non-seeded specimen to
verify that the behaviour exhibited in the seeded
case is reasonable, and the effect of the seeding
FIGURE 4. Rod impacting glass at 190 m s"1, 105 |is delay.
805
REFERENCES
1.
2.
3.
4.
100
200
300
400
500
600
700
800
pixels
FIGURE 6. Displacement field for rod impact after 40 us.
does not appear to have altered the structural
response of the glass significantly. The natural
progression now is to make thicker, and hence more
realistic glass samples. These samples can then be
used to measure the displacement field at varying
delay times, projectile velocities and depths through
the sample. The full stereoscopic X-ray speckle
treatment can also then be applied to give a truly
three-dimensional measure of the internal
displacements4.
ACKNOWLEDGEMENTS
The authors thank Prof. I.E. Field and Dr H.T.
Goldrein, Cambridge and Dr. I.G. Cullis (DERA)
for their advice and encouragement. The research is
supported by the Engineering and Physical
Research Council (EPSRC), and the Defence
Evaluation and Research Agency (DERA). R.
Smith is thanked for his technical help in preparing
the samples.
806
Synnergren, P., Goldrein, H.T., Proud, W.G., Appl.
Opt. 38, 4030-4036 (1999).
Bourne, N.K., Forde, L.C., Millett, J.C.F., Field, J.F.,
/. Phys. IV France Colloq. C3 7, 157-162 (1997).
Sjodahl, M., Benckert, L.R., Appl. Opt. 32, 22782284 (1993).
Goldrein, H.T., Synnergren, P., Proud, W.G.,
"Three-Dimensional Displacement Measurements
Ahead of a Projectile," in Shock Compression of
Condensed Matter-1999, edited by M.D. Furnish,
L.C. Chhabildas, and R.S. Hixson, AIP Conference
Proceedings 505, Snowbird, Utah, 1999, pp. 10951098.